Skip to main content

Early Screening for Autism Spectrum Disorders in Health Care Settings

Final Report Summary - ASD SCREENING (Early Screening for Autism Spectrum Disorders in Health Care Settings)

Project context and objectives

This study evolved in an era in which screening for autism spectrum disorders (ASD) has been questioned leading to an intense scientific effort to gain a better understanding of what to screen for, when, how, and using what risk criteria. Researchers across the world are trying to understand the precursors and development of ASD. Although the rate of ASD diagnosis is on the rise internationally, and the state-of-the-art practice parameters for ASD endorse broad population screening, routine universal screening for ASD is not mandated in most countries. This is due in part to cost and in part to lack of evidence validating ASD-screening tools for children under the age of two years. Currently, a majority of children are not diagnosed with ASD until the second and third year of life or later, although if probed, symptoms can be observed earlier. The primary incentive for early identification is for promoting access to earlier intensive autism-specific interventions. Such interventions can improve a child's prognosis, especially in developing communication and functional skills, and in managing behaviour. Since the only means for screening and diagnosing ASD are observational, it is imperative that we develop valid and reliable screening tools.

The overall goal of this screening project was to study the efficiency of early screening tools for ASD by weighing their cost versus benefit for society. This goal has been addressed through the close examination of the feasibility, reliability, and validity of ASD screening at 12 months, within community day-care and healthcare settings. Multiple components of screening procedures for ASD have been investigated in this project, including aspects such as the prediction of ASD risk from 12 to 30 months, the comparison of risk outcomes across different measures and relative to other screening procedures in the community. Our second goal was to study the development of sensory over-responsivity in the general population and in children already diagnosed with ASD, and how it relates to their anxiety.

Project methods

To achieve the first goal, three subsamples of 12-month infants were screened. 613 infants from day-care centres were screened using the full measures of first year inventory (FYI), 583 infants from baby-wellness clinics were screened with 'FYI-Lite' procedures, and 82 infants, for comparing father and mother risk, report on the full version. A subset of infants was evaluated closely at home using clinical observation measures including the autism observation scale for infants (AOSI) and the Mullen early learning scales (MSEL). This subset comparison of infants meeting on various types of risk, namely social, sensory and social, mean risk, socio-economic status (SES) and gender, matched the control group. This subset was also followed at 30 months to determine diagnostic outcome, using the ADOS and MSEL as well as standardised parent-report instruments. Based on these assessments infants were referred at 13 and/or at 30 months. Observations were conducted by a team of expert clinicians blind to the child’s risk status on the FYI.

In the area of sensory over-responsivity, we used existing data from a birth cohort followed from infancy to school age as well as a sample of toddlers with ASD followed for three years with a battery of child and parent measures. Growth-curve latent models were applied to study growth trajectories, associations and predictions over time and between different areas of impairment.

Main project results

The main results were that the optimal sensitivity and specificity rates for the FYI were achieved once both threshold and type of ASD marker - sensory and social - are taken into consideration. Infants with sensory and social risk at 12 months showed lower MSEL developmental scores and higher AOSI risk scores as well as a higher rate of persisting referral at both time points relative to those with no risk or risk in only one domain. We found that for most FYI items, norms developed in the US were applicable to the Israeli sample, however we identified a few items that were skewed in their distribution and may not be appropriate for screening for ASD in Israel. Within the day-care sample a total of five infants were identified with ASD at 30 months of age.

The baby-wellness-clinic sample risk scores were compared to their provider routine-screening outcome, enabling us to compare the outcome for all screened infants and not only those infants who we observed. FYI-L risk status relative to the outcome of clinic providers' routine screening indicated 41.67 % agreement on positive risk. Exploratory analyses were conducted to examine whether age differences and/or the rate of preterm births can explain 'false' FYI-L results relative to routine screening. An analysis of variance, comparing age at outcome between true positive (M = 22.20 months, 16 - 26), true negative (M = 21.06 months, 14 - 28), false positive (M = 20.71 months, 15 - 26) and false negative (M= 24.40 months, 20 - 27) infants, was significant (F (3, 491) = 3.13 p = 0.03). Tukey post-hoc tests showed that the false negative subgroup was older than the true negative subgroup (p = 0.02). Chi square tests showed significant differences in the frequency of preterm births (2 (3) = 18.41 p < 0.001) with 60 % of those with true-positive risk results being born preterm while less than 10 % in the other subgroups.

Several SES factors were associated with greater risk, such as parent lower level of education and single parenting. The matching of control and risk groups on SES allowed us to look at developmental differences above and beyond socio-economic disparities.

The cross-informant reliability subsample of 82 fathers and mothers that completed the FYI showed that parents highly agreed on the sensory-regulatory risk status of their child, however fathers tended to report higher risk on the social-communication domain. Internal reliability was good, as well as validity compared with the Infant-Toddler Social and Emotional Assessment (ITSEA) questionnaire.

The main sensory over-responsivity results were that sensitivity early on predicts later over-responsivity in school-age children, however within the three first years of life is stable. In addition we found that sensitivity does not change in children with ASD over the first three years following diagnosis but predicts changes in anxiety within these years.

Project impact

Screening results have great impact on society by highlighting the importance of specific ASD screening in infancy. The diverse sample screened provides insight into the unique informant perspective and developmental status of infants from families with fewer resources. The study also shows that risk outcome for ASD based on father reporting is not comparable to that of mothers, and separate norms need to be developed for father-based screening. The study also shows that a follow-up on a parent-screening tool is critical for reducing false referral to clinical evaluation, avoiding overload on family and healthcare systems. In addition, our study indicated that not all infants later diagnosed with ASD will be identified at 12 months, as there are different types of courses for ASD symptoms, with some showing sub-threshold markers at 12 months and others not showing any risk at all at 12 months. These different patterns of ASD development may correspond to different neurogenetic pathways to be further explored. In addition, our sensory over-responsivity findings support the long-lasting question of a causal relation between sensory sensitivity and anxiety in ASD, by demonstrating that early sensory sensitivity predicts changes in anxiety over time.

This project has targeted multiple groups leading to a great increase in awareness and training for early ASD screening in the community. These groups include day-care administrators, directors and staff, parents of infants, health administrators and healthcare providers at baby-wellness clinics. In addition, the various presentations and publications on scientific platforms enabled us to target students, paraprofessionals and researchers, and disseminate the outcomes of the study and its relevance to research and clinical needs in this area.